For many years typewriters and most other data entry and control devices have used versions of the standard (QWERTY) keyboard which appeared on the first typewriters. There have been minor changes in the location of keys and the numbers of rows and columns. Symbols, functional controls, etc. have been added as keyboards were adapted to new types of systems and equipment. The pace of change in keyboards has accelerated with the introduction of computers and the development of new kinds of switching devices; matrix key switches, Hall elements, etc. However, most new keyboards have retained the basic features of the Sholes typewriter keyboard designed in 1873:
1. selection of characters by operation of one key at a time; and PA1 2. layout of keys in two or more rows for each hand. PA1 "The significance of the present results should be examined both from an applied and a theoretical viewpoint. Taken together, they showed that: (a) subjects were able to perform in a touch typing mode, and memorize the codes for all letters, after a brief period of self teaching, (b) that progress in learning was fast and included the development of parallel entry capabilities, (c) that in general, representation of codes by spatial patterns was considerably better than coding by hand symmetry, (d) that performance with an upright tilted panel was better than with a horizontal panel. PA1 These findings raise the possibility that for many system applications, a chord keyboard of the type described in the present study may constitute a viable and attractive alternative to the traditional typewriter or data entry keyboard. Touch typing ability and similar entry speeds on a standard keyboard are the achievements of several months of daily practice (e.g., Hill, Rejall & Thorndyke, 1913; Dvorak, Merrick, Dealy & Ford, 1936; Gentner, 1982). In light of the fundamental differences between the two typing keyboards in their skill components, one can conclude that it appears easier for humans to commit 52 chords to memory and activate them upon request, than to learn the ways of the hand to a similar number of keys spread out on a typing keyboard. This conclusion is also supported by another line of experiment, with a single hand chord typewriter for the Hebrew language (Gopher & Eilam, 1979; Gopher, in press)"
There is another type of keyboard which has found limited use; the chord-type keyboard. In a chord-type keyboard entries are made by operating two or more keys simultaneously. With as few as 10 keys the entire set of numbers and letters and a large number of control characters may be encoded using combinations of two or more keys. There are, in fact, 1024 possible states of 10 keys, seemingly more than enough to represent any desired character set. Despite the large number of states available, 10 keys are not enough for reliable data entry, however. This is because the transitions from state to state are as important as the individual states themselves. When assigning keyboard states to characters, the designer of a chord keyboard must try to avoid assigning a character to any state which is liable to occur by accident between any two valid states. This constraint is an important one, because during entry of text or other data, any sequence of two valid characters may occur. Premature release or early operation of one of the keys in a combination is extremely likely to result in an unwanted entry. As the operator's skill and speed increase, this problem becomes even more severe. Anyone who has operated a chord-type keyboard with ten or fewer keys can testify to the necessity for extreme care and precise timing. The problem of unwanted combinations occurring during transitions is likely to have been the most significant deterrent to adoption and use of chord-type keyboards with 10 or fewer keys.
It has been demonstrated that memorization of chord key combinations to represent letters, numbers and a few control characters is a task readily accomplished by keyboard operator trainees. An extensive 1983 study of two-hand chording by D. Gopher and W. Koenig ("Hands Coordination in Data Entry With a Two Hand Chord Typewriter", Report CPL-83-3, Contract N00014-83-K-0092, Illinois University Cognitive Psychophysiology Laboratory, Champaign, Ill.), resulted in the following conclusions:
It has also been demonstrated experimentally by Ratz, H. C., and Ritchie, D. K. ("Operator Performance on a Chord Keyboard" Journal of Applied Psychology, Vol. 45, No. 5, 1961, 303-308), that subjects who are shown a visual representation of a chord pattern, and then are required to reproduce that pattern in a keyboard with the fingers of one hand, respond most rapidly to single finger patterns. Responses to multiple-finger pattern stimuli take as much as 50% longer.
There have been many attempts to design chord-type keyboards for data entry or machine control, but only two have achieved noteworthy success. The Stenotype machine, a product of the Stenograph Corporation, uses chording to represent a shorthand language for rapid transcription of speech in courtrooms and other legal proceedings. While the operator of this machine presses more than one key at a time to represent a word or phrase, the machine itself makes a separate record of each key depression. Thus chording is not so much a function of the machine as it is of the shorthand language. See U.S. Pat. No. 2,393,781, "Stenographic Machine" by C. W. Johnson, J. G. Sterling, M. H. Wright and R. T. Wright for a description of how this device operates.
The United States Postal Service has employed chording in mailbag dispatching systems, using the chord combinations to represent extraction or destination codes (see Cornog, J. R., Hockman, J. F. and Craig, J. C. "Address Encoding--A Study of the Double-binary Keyboard as a Link in the Machine Sorting of Mail" ASME 63-WA-338, Paper presented to the American Society of Mechanical Engineers, New York 1963).
A number of chord-type keyboards have been patented, however none of these has achieved wide acceptance in the marketplace. An early, simple type of chording is illustrated by R. R. Seeber's "Word Writing Typewriter" of U.S. Pat. No. 2,717,686. This machine has a special key which transforms the typewriter keyboard into a word or phrase generator (anticipating today's "Smartkey" computer keyboard enhancement software).
R. Seibel et al described a one hand chord-type "Communication Device" in U.S. Pat. No. 3,022,878. The problem which was alluded to above, of unwanted characters being produced by transition states, is illustrated well by FIG. 6 of this patent, a portion of which is represented below:
______________________________________ ##STR1##
______________________________________
Here the character "H" is represented by the index finger being gripped (g) and the middle finger being extended (e). If the next character to be entered is a "G", the positions of these two fingers must be reversed, i.e. the middle finger must be gripped and the index finger extended. Any of four valid characters may be accidently produced during this transition. If the middle finger lags behind the index in leaving the "H" position, an "I" will be produced. If the index finger lags, an "M" will be produced. If the middle finger lags in reaching the new "G" position, a "U" will be produced. If the index finger lags in reaching the "G" position, a "," will be produced. Thorough examination of FIG. 6 will reveal that there are very few transitions free of this problem. The solution adopted by Seibel is to inhibit recognition of any new character until a delay, started by the first switch closure, has time out. The delay must be long enough to inhibit errors with the normal dispersion of operate times, but not long enough to slow the entry process.
In U.S. Pat. No. 4,042,777, "One-handed Keyboard and its Control Means", F. C. Bequaert et al address the transition error problem by adding a circuit which recognizes a character only when the first key of a chord is released. The inventors further reduce the probability of unwanted characters during transitions by using a more restricted set of combinations; i.e. only those produced by adjacent keys.
In his U.S. Pat. No. 4,344,069, "Method and Apparatus for Character Generation", E. S. Prame carries the use of key time sequence one step further, distinguishing between characters encoded by single keys and those represented by chords, with a sequential switching circuit. The operator of such a keyboard may not overlap key strokes for single-key characters, or chord characters will be produced by accident. A similar technique to that of Prame was used earlier in a telephone-style keyboard by R. W. Conway and H. L. Morgan in their "Tele-CUPL: A Telephone Time Sharing System," reported in Communications of the ACM, Volume 10, Number 9, September 1967 (Pp. 538-542).
Another type of keyboard which appears in patents and other publications is one using multiple position keys. The "Electrically Operated Typewriter" described by F. H. Hesh in U.S. Pat. No. 2,532,228 is an example. This machine has only ten keys, but each key has four active positions. Any of 40 characters may be printed by moving a key individually to one of its four positions. R. A. Samuel later patented an "Electrical Typewriter Key and Keyboard Arrangement" (U.S. Pat. No. 3,633,724) which uses eight five-position keys to select the same number of characters in a similar fashion.
The "Digital `X` Keyboard" described by D. L. Conway in International Business Machines Corporation, Technical Disclosure Bulletin Vol. 18, No. 12 of May 1976 (Pp. 4187-4190) also uses multiple state keys, relying on an experienced typist's familiarity with the QWERTY keyboard layout to be able to "touch" type, a single key at a time, with little training.
Multiple position key configurations for such machines are described by W. Zapp in two U.S. Pat. Nos.: 4,081,068 and 4,201,489, both entitled "Keyboard Actuable With the Aid of the Fingers of at Least One Hand".
The motor skills needed to operate any of the keyboards mentioned above are neither rare nor hard to acquire. Apparently the task of memorizing up to 100 chord combinations is within the capability of most operator trainees. Why, then, have those mechanically simple keyboards not replaced the venerable QWERTY? The answer may be that with all keyboards developed to date which have ten or fewer keys, the demands for precision of motion and timing are too great, and an excessively long training period is required to build speed and accuracy.